Some time this morning (just before 8 judging from syslog etc) my “new” server (ok I’ve had it about a year now and it was a medium-sized ISP’s primary DNS server before that) went pop. Well, at least that’s my best approximation of it, I think the PSU has gone but I don’t have a spare that will fit it.
Luckily, I’d not got rid of my old desktop (I was going to put OpenSolaris on it at some point) or the old server, so I grabbed the old network card from the old crusty server (Celeron or something, can’t even remember) and stuck it in the old desktop, along with the hard drive from my now dead server.
At first it seemed to boot up nicely, well at least it seemed to start booting up, then got stuck at “waiting for root filesystem” - which worried me somewhat. I booted the box again, saw that it picked up the hard drive as hda, then noticed that grub was pointing to /dev/hdc1 - AHA!
Quick reboot and edit of grub options later, and I was in single user mode with the disk showing correctly. A few fsck’s later I was able to edit the grub config permanently and sort out the fstab.
Another reboot later (to get into multiuser) and the network wasn’t up, how odd. I did a quick ifconfig and the interfaces were now eth2 and eth3 instead of eth0 and eth1! Fortunately I remembered something from work about udev in Debian etch and that it can be configured, so after a quick bit of grepping I found the relevant file (
/etc/udev/rules.d/z25_persistent-net.rules), ripped out the eth0 and eth1 lines and edited the remaining ones accordingly.
After yet another reboot, it was all up and running, and I could relax a bit. Unfortunately for me the server only has 512MB RAM and an Athlon 1GHz CPU, as apposed to the dead server’s 1.8GHz Athlon and 2.5GB RAM. Oh well, at least my connection is back!